Influence of Project Type and Procurement Method on Rework Costs in Building Construction Projects

Influence of Project Type and Procurement Method on Rework Costs in Building Construction Projects Peter E. D. Love 1 Abstract: While it is widely recognized that additional costs due to rework can have an adverse effect on project performance, limited empirical research has been done to investigate the influencing factors. The research presented in this paper aims to determine the influence of different project types and procurement methods on rework costs in construction projects. Using a questionnaire survey, rework costs were obtained from 161 Australian construction projects. The direct and indirect consequences of rework are analyzed and discussed. It is shown that, contrary to expectation, rework costs do not differ relative to project type or procurement method. In addition, it was found rework contributed to 52% of a project s cost growth and that 26% of the variance in cost growth was attributable to changes due to direct rework. To reduce rework costs and therefore improve project performance, it is posited that construction organizations begin to consider and measure them, so that an understanding of their magnitude can be captured, root causes identified, and effective prevention strategies implemented. DOI: 10.1061/ ASCE 0733-9364 2002 128:1 18 CE Database keywords: Project management; Costs; Australia; Scheduling. Introduction The doctrine of time is money appears to be well entrenched in the psyche of almost all construction industry clients. To satisfy the requirement of time completion on time or earlier a plethora of nontraditional procurement methods have emerged in the marketplace, which has resulted in design and construction schedules being compressed and construction commencing before the final design is complete Hanna et al. 1999. As design and construction time is compressed, the degree of overlap which will be referred to as concurrency between activities increases, which in turn increases project complexity as activities are subdivided into trade packages. Hoedemaker et al. 1999 suggest that there is a limit to the number of tasks that can be undertaken in a concurrent manner. Beyond this specified limit, the probability of rework occurring as well as time and cost overruns being experienced significantly increases. Primarily this is due to the complexities associated with communication and coordination of a large number of tasks undertaken in a concurrent manner Love et al. 2000a. The problems associated with concurrency are exacerbated by time pressure, especially when rework occurs, and can accumulate as the data of project completion is approached. In an effort to meet the project s scheduled completion date, design and construction firms may have to employ additional resources. Such actions may lead to the opposite of the desired effect. That is, 1 Associate Professor, School of Management Information Systems, Edith Cowan Univ., Churchlands, Perth, WA 6018, Australia. E-mail: p.love@ecu.edu.au Note. Discussion open until July 1, 2002. Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and possible publication on January 23, 2001; approved on March 6, 2001. This paper is part of the Journal of Construction Engineering and Management, Vol. 128, No. 1, February 1, 2002. ASCE, ISSN 0733-9364/2002/1-18 29/$8.00 $.50 per page. pushing beyond the limits of effective concurrency increases complexity, and this can increase the time to complete tasks Brookes 1975. Considering these issues, nontraditional methods may in fact be subject to higher rework levels than traditional methods, especially when errors, omissions, and/or changes occur. Moreover, under traditional methods, design and documentation are supposed to be complete, or largely complete, before construction commences on-site, so in theory there should be less rework attributable to design-related sources. With this notion in mind, the primary objective of the research reported in this paper is to determine the influence different procurement methods and project types have on rework costs in projects. Using a questionnaire survey, rework costs from 161 Australian construction projects were obtained. From this survey, rework costs and their impact on project cost and schedule are analyzed and discussed. While the research was conducted in an Australian context, it is envisaged that the research outcome would be widely applicable in other locations. Prior to the presentation of the research, a brief review of the costs of rework in projects is presented. Rework Costs Various interpretations of rework can be found in the construction management literature. For example, terms such as quality deviations Burati et al. 1992, nonconformances Abdul-Rahman 1995, defects Josephson and Hammarlund 1999, and quality failures Barber et al. 2000 are often used, though these definitions vary. Ashford 1992 defines rework as the process by which an item is made to conform to the original requirement by completion or correction. The Construction Industry Development Agency 1995, however, defined rework as doing something at least one extra time due to nonconformance to requirements. Essentially, rework can result from errors, omissions, failures, damage, and change orders throughout the procurement process Love et al. 1999; Love and Li 2000b. 18 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002

Josephson and Hammarlund 1999 reported that the costs of residential, industrial, and commercial building projects range from 2 to 6% of their contract values. Similarly, Love and Li 2000b in their study of rework costs for a residential and industrial building found the costs of rework to be 3.15 and 2.40% of contract value, respectively. In addition, Love and Li 2000a found that when a contractor implemented a quality assurance system in conjunction with an effective continuous improvement strategy, rework costs were found to be less than 1% of the contract value. The costs of quality deviations in civil and heavy industrial engineering projects, however, have been found to be significantly higher. Burati et al. 1992 studied nine major engineering projects to determine the cost associated with correcting deviations to meet specified requirements. The results of their study indicated that, for all nine projects, quality deviations accounted for an average of 12.4% of the contract value. A significantly lower figure was reported by Abdul-Rahman 1995, who found nonconformance costs excluding material wastage and head office overheads in a highway project to be 5% of the contract value. Abdul-Rahman 1995 specifically makes the point that the nonconformance costs may be significantly higher in projects where poor quality management is implemented. Notably, Nylén 1996 found that when poor quality management practices were implemented in a railway project, quality failures were found to be 10% of the contract value. Nylén 1996 further found that 10% of the quality failures experienced accounted for 90% of their total cost. Here, significant proportions 76% of the quality failures were attributable to design-related issues, such as erroneous documentation and poor communication between project team members. As mentioned above, rework can also originate from change orders Knocke 1993; Love and Li 2000b. However, the extent to which change orders contribute to rework costs remains relatively unexplored. Research undertaken by Zeitoun and Oberlander 1993 found that the median costs of change orders for 71 fixedprice projects were 5.3% of the contract value and 6.8% for 35 cost reimbursable projects. Similarly, research undertaken by Cox et al. 1999 in the U.K. revealed that the costs of design-related change orders could range from 5 to 8% of the contract value, even when projects are managed effectively, as most of the changes are initiated by clients. The costs of change orders in the research reported by Zeitoun and Oberlander 1993 and Cox et al. 1999 are similar to the rework costs previously reported. A degree of change can be, and to a certain extent, should be expected in construction, as it is difficult for clients to visualize the end product that they procure. However, almost all forms of rework with the exception of that caused by weather are preventable, since poor management of the design and construction process typically causes such costs to occur. Research Methodology The research presented in this paper is part of a wider study that sought to determine the influence that project attributes, organizational management practices, and project management practices have on rework costs in construction projects. For the purposes of the research presented in this paper, rework is defined as the unnecessary effort of redoing a process or activity that was incorrectly implemented the first time. Rather than developing a questionnaire survey that sought respondents general opinions about rework, the writer asked respondents to select a recently completed project most familiar to them and to subsequently answer questions about the perceived causes of rework, associated costs, and project management practices implemented. In essence, each project was treated as a separate case. In examining rework costs experienced in the projects sampled, the following two hypotheses were addressed: 1. There are significant differences in rework costs with varying procurement methods for construction projects; and 2. There are significant differences in rework costs for varying types of construction projects. Questionnaire Survey Fig. 1. Respondents by profession Stratified random sampling was used to select the study sample from the telephone directory Yellow Pages. In addition to increasing the representativeness of samples, stratified random sampling was a useful technique that made general statements about the portions of the population possible. Prior to determining the sample size for the main study, a pilot survey was undertaken with 30 selected firms, which comprised architects, project managers, and contractors from the Geelong and Melbourne region, in the state of Victoria. This was undertaken to test the potential response rate, suitability, and comprehensibility of the questionnaire. Each firm was contacted by telephone and informed of the aims of the research. On obtaining their consent, the questionnaire was mailed, with a stamped addressed return envelope enclosed, for respondents returns, comments, feedback, and completion. The respondents were also asked to critically review the design and structure of the survey. All comments received were positive, and as a result the questionnaire remained unaltered for the main survey. A total of 25 responses were received in the pilot survey, which equates to a response rate of 83%. The composition of respondents that returned the questionnaire was architects 30%, contractors 50%, and project managers 20%. In the main survey, 70 questionnaires were mailed to each of the aforementioned categories of respondents throughout Australia, which equates to a total of 420 questionnaires distributed. One hundred thirty-six valid responses were received from the main survey. As the pilot questionnaires required no changes, they were added to the sample, which resulted in 161 valid responses representing a total consolidated response rate of 36%. This response rate is considered acceptable for a survey focusing on gaining responses from industry practitioners Alreck and Settle 1985. Fig. 1 provides a breakdown of the valid responses by respondent type. Contractors, architects, and consultant project managers accounted for approximately 81% of the respondents. While JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002 / 19

Fig. 2. Respondents by state quantity surveyors QSs and structural, mechanical, and electrical engineers appear to be underrepresented; it should be noted that many consultancy firms offer project management services and as a result may have undertaken a role of project manager for a project that they selected. As the response rates from the QSs and the engineering profession were low, the respondents were recategorized under the following headings for analysis purposes: Design consultants, which comprised architects, QSs, and structural, mechanical, and electrical engineers 44% ; Contractors 33% ; and Consultant project managers 23%. Number of stories, Procurement method, Tendering method, Original contract value, Contract value on practical completion, Original construction period, and Actual construction period. In addition, estimates of direct and indirect rework costs that occurred during the project s construction were also sought. Procurement and tendering methods and project and facility types were measured on a nominal scale, while the other variables were measured on a ratio scale. Respondents were also asked to provide general comments at the end of the questionnaire with respect to the sources and causes of rework in their selected project. The data collected were analyzed using the Statistical Package for the Social Sciences for Windows, Version 9.00. Descriptive statistics were used to determine the central tendency and dispersion mean and standard deviation of the data. Such descriptive analysis provided a frame of reference for using appropriate inferential statistical techniques. In order to test the associations of the variables that were measured on a ratio scale with the rework costs, Pearson s parametric correlation was computed, and this approach enabled those variables with significant correlation at the 95 and 99% significance intervals to be identified. As well as determining the relationships between the variables, a one-way analysis of the variance ANOVA was used to test for significant differences between rework costs for different procurement methods and project types. Moreover, to identify where any differences between samples lie, a Tukey s post hoc test was used. Data Reliability Data reliability is related to data source and the identification of the position held by the person who completed the questionnaire Oppenheim 1992. Thus, as it was important that the personnel who had detailed knowledge about the procurement processes associated with a project answered the questionnaire, it was mailed to the senior personnel within the organizations identified. Whether they actually answered the questionnaire was impossible to determine, except in circumstances where respondents optionally supplied their details at the end of the questionnaire. From the 161 questionnaires used in the research, 87 respondents provided their business details, and many held senior positions within their organizations. Based on the position titles of the respondents, the direct mailing to individuals in organizations seemed to have achieved its objective of reaching those who were closely involved with delivering construction projects. In addition, the questionnaires were mailed to organizations in different states in Australia, which minimized the duplication of selected projects. Fig. 2 provides a breakdown by state of the respondents who answered the questionnaire. Considering the number of construction projects being undertaken in Australia at any given time, the likelihood of the respondents selecting the same project was significantly reduced because of the diversity of data sources from each state. Method and Data Analysis Justification The respondents were asked to provide the following details: Project and facility type, Gross floor area GFA, Analysis and Discussion Project and Facility Type A variety of project and facility types and procurement and tendering methods were identified in the survey. Tables 1 and 2 present a summary of project and facility types and the respective procurement methods used to deliver projects obtained from the survey data. Table 1 shows that traditional lump sum methods were typically used to procure new build 44.4% and refurbishment projects 72%, whereas construction management methods tended to be used to procure fit-out projects 42.9%. New build project types accounted for 56% 90 of the total number of projects Table 2. Refurbishment and renovation projects were found to be the second most popular type, accounting for 26% 43 of the total projects. Table 2 indicates that respondents were involved in the procurement of a variety of facility types. The most popular facility types were commercial offices 13.7%, residential buildings 13%, and hospitals/health 12.4%. Procurement and Tendering Methods From the sample of 161 projects, the most popular procurement methods used to deliver project types were traditional lump sum 52.2%, design and build 19.9%, and construction management 16.1%. Previous research undertaken by Love et al. 1998b, which examined procurement selection among clients and consultants, found that these forms of procurement methods were also predominant in the marketplace. Similarly, traditional lump sum methods are also the most popular forms of procurement in many other Commonwealth countries such as Malaysia Hashim 20 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002

Table 1. Procurement Methods Used to Deliver Project Types Procurement method Project type Refurbishment/ New build/ New build renovation fit-out refurbishment Combination of all Total Number % Number % Number % Number % Number % Number % Traditional lump sum 40 44.4 31 72.0 5 35.7 7 63.6 1 33.3 84 52.2 Traditional cost plus 1 1.1 1 0.6 Traditional with provisional quantities 5 5.6 5 3.1 Design and manage 1 1.1 1 7.1 2 1.2 Construction management 11 12.2 6 14.0 6 42.9 3 27.3 26 16.1 Management contracting 1 1.1 1 0.6 Design and build 22 24.4 6 14.0 2 14.3 1 9.1 1 33.3 32 19.9 Novation 5 5.6 1 33.3 6 3.7 Turnkey/package deal 2 2.2 2 1.2 Build-Own-Operate Transfer BOOT 2 2.2 2 1.2 Total 90 100 43 100 14 100 11 100 3 100 161 100 1997, Hong Kong Chan et al. 1999, Singapore Lam and Chan 1995, South Africa Bowen et al. 1997 and the U.K. RICS 1996. Traditional methods of procurement have been heavily criticized for their sequential approach to project delivery, as they have contributed to the so-called procurement gap whereby design and construction processes are separated from one another Love et al. 1998a. As a result, Love et al. 1998a suggest that behavioral, cultural, and organizational differences between project individuals and groups often prevail. In addition, the procurement gap between design and construction is considered to inhibit communication, coordination, and integration among project team members and can adversely affect project performance Lahdenperä 1995; Evbuomwan and Anumba 1996. Nontraditional methods such as design and build and construction management have been advocated as methods for overcoming some of the problems inherent in traditional methods NEDO 1988; Turner 1990; Masterman 1994, yet it would appear from these findings that their use is minimal Table 2. Sharif and Morledge 1997 provide a plausible explanation for the ubiquitous use of traditional methods by stating that most construction clients are small and occasional and therefore only ever build once or twice. Thus, an architect is typically the first point of contact for these clients and their advice is heavily relied upon Mackinder and Marvin 1982; Sharif and Morledge 1997. Consequently, it is often in the interest of the architect to persuade the client to use a traditional method, as they can take a lead role in the project as well as maximize his or her fees. Traditional methods can provide clients with cost certainty, whereas design and build and construct management methods are often used when the pressure of early completion is imposed on the project Holt et al. 2000. Table 3 indicates that single-stage tendering was the most popular tendering method 52%, then negotiation 32%, and finally two-stage tendering 16%. No other forms of tendering were identified as being used in the procurement of the 161 projects. Notably, single-stage tendering was predominantly used in conjunction with traditional methods, whereas negotiated types tended to be used more with nontraditional methods such as de- Table 2. Type of Facility and Project Procured Project New build Refurbishment/renovation Fit-out New build/refurbishment Combination of all Total type Facility type Number % Number % Number % Number % Number % Number % Administrative authorities 3 3.3 3 7.0 2 14.3 8 5 Administrative civic 8 8.9 2 4.7 2 14.3 12 7.5 Banks 1 1.1 2 4.7 2 18.2 5 3.1 Educational school 3 3.3 2 4.7 5 3.1 Educational university 6 6.7 3 7.0 2 18.2 1 33.3 12 7.5 Entertainment 5 5.6 1 2.3 1 7.1 7 4.3 Hotel/motel/resort 11 12.2 1 2.3 1 7.1 13 8.1 Hospitals/Health 10 11.1 4 9.3 1 7.1 5 45.5 20 12.4 Commercial recreational 9 10 2 4.7 1 7.1 12 7.5 Commercial retail 7 7.8 5 11.6 2 14.3 1 9.1 1 33.3 16 9.9 Commercial offices 8 8.9 9 20.9 3 21.4 1 9.1 1 33.3 22 13.7 Industrial warehouses 1 2.3 1 7.1 2 1.2 Industrial factory 2 2.2 2 4.7 4 2.5 Residential 16 17.8 5 11.6 21 13 Airport 1 1.1 1 2.3 2 1.2 Total 90 100 43 100 14 100 11 100 3 100 161 100 JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002 / 21

Table 3. Tendering and Procurement Methods Used to Deliver Project Types Tender method Single stage Two-stage Negotiated Total Procurement method Number % Number % Number % Number % Traditional lump sum 57 68.7 12 44 15 29.4 84 52.2 Traditional cost plus 1 1.2 1 0.6 Traditional with provisional quantities 1 1.2 1 3.7 3 5.9 5 3.1 Design and manage 2 2.4 2 1.2 Construction management 10 12 4 14.8 12 23.5 26 16.1 Management contracting 1 3.7 1 0.6 Design and build 8 9.6 7 25.9 17 33.3 32 19.9 Novation 3 3.6 1 3.7 2 3.9 6 3.7 Turnkey/package deal 1 3.7 1 2.0 2 1.2 BOOT 1 1.2 1 2.0 2 1.2 Total 83 100 27 100 51 100 161 100 sign and build and construction management procurement methods. In the U.K., however, the most popular methods of tendering, in order of popularity, have been found to be selective tendering, negotiation, and two-stage and open tendering Holt et al. 2000. In an earlier study in the U.K., Bresnen et al. 1988 found that a quarter of contacts were negotiated, which is surprising in that negotiation removes the element of competition. Nonetheless, this correlates with the findings of the then Institute of Building 1979, who stated Competition is not essential to achieve value for money. Evidence shows that contracts let by open or single stage selection appear less successful than those let by other means whilst negotiated contracts show less divergence between final, and tender values. Whether there is actually less divergence between final and tender values when using negotiated forms of tendering will be explored using the ANOVA test. Thus, cost growth the divergence between original and actual contract value on completion was identified as the dependent variable and tendering type the independent variable. Levene s test of homogeneity of variances was not violated (p 0.05), which indicates that the population variances for tendering method were equal. The ANOVA revealed no significant differences between each tendering method for cost growth, F(2, 158) 0.631,p 0.05. Selective competition places emphasis on lowest cost, and when Bills of Quantities are included, cost certainty and quality can be achieved Bresnen et al. 1988. However, according to Watkinson 1992, traditional methods tend to perform poorly in terms of time. In stark contrast, however, the findings reported above clearly indicate that there is no difference between procurement methods in terms of their schedule growth. Accordingly, Walker 1994 also found that procurement methods do not influence the time performance of projects. project with a contract value of $A25,500 to 238,000 m 2 for a new build airport terminal with a contract value of $A390,000,000 M 13,583 m 2, SD 23,617 m 2. Surprisingly, the fit-out project was estimated to have experienced rework costs of 12% 10% direct plus 2% indirect, whereas the airport terminal experienced only 2% rework costs direct only. In determining whether there was a significant relationship between GFA and rework costs, Pearson s r correlation was computed Table 4. The correlation for the data revealed that GFA and number of storeys were not significantly related to rework costs. Cost and Schedule Growth Construction projects are notorious for running over budget and time Hester et al. 1991; Zeitoun and Oberlander 1993; Ibbs and Allen 1995. Change orders have been found to be a major contributor to time and cost overruns Jahren and Ashe 1990, yet the impact that rework has on the cost and time performance of projects remains unexplored in the construction management literature. Hence, respondents were asked to provide the following details so that cost and schedule growth could be calculated for each project: Project Size Fig. 3 identifies the number of projects by contract value. The contract values for the projects were found to range in Australian dollars from $A25,500 to $A390,000,000 mean (M) $A25,521,927, standard deviation (SD) $A51,957,899. Likewise, the contract duration ranged from 2 weeks for a retail fit-out to 450 weeks for an airport terminal M 66 weeks, SD 46 weeks. The gross floor area GFA for the 161 projects used in the analysis was found to range from 116 m 2 for a fit-out Fig. 3. Number of projects by contract value 22 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002

Table 4. Correlation Matrix of Rework Costs and Project Characteristics Project characteristics 1 2 3 4 5 6 7 8 Cost growth 1.00 Gross floor area 0.00 1.00 Number of floors 0.09 0.32 b 1.00 Rework as percentage of cost growth 0.00 0.12 0.12 1.00 Schedule growth 0.31 b 0.12 0.08 0.15 a 1.00 Total rework cost 0.44 b 0.05 0.01 0.15 a 0.38 b 1.00 Direct rework cost 0.51 b 0.08 0.05 0.20 a 0.38 b 0.91 b 1.00 Indirect cost of rework 0.27 b 0.01 0.07 0.07 0.31 b 0.89 b 0.63 b 1.00 a Correlation is significant at 0.05 level 2-tailed. b Correlation is significant at 0.01 level 2-tailed. Original contract value, Project s expected duration in weeks, Actual contract value on practical completion, and Actual construction period. Cost and schedule growth for each project was calculated using the following formula Zeitoun and Oberlander 1993 : PROJECT CG CVP OCV OCV (1) PROJECT SG ACP OCP (2) OCP where CG percentage cost growth; SG percentage schedule growth; CVP contract value on practical completion; OCV original contract value; ACP actual construction period weeks ; and OCP original construction period weeks. Table 5 displays the mean cost and schedule growth experienced for projects delivered using different procurement methods. Figs. 4 and 5 denote the aggregate distributions of cost ( M 12.6%,SD 24.22%) and schedule growth (M 20.7%,SD 28%) for the 161 projects. The maximum cost growth was 244% and the minimum 84%, which results in a range of 328%. Moreover, the maximum schedule growth was 144% and the minimum 29%, which results in a range of 170%. It is evident from the findings presented above that cost and schedule overruns in projects were a common occurrence and that the degree of overrun varied significantly among projects. Nevertheless, is there a difference between procurement methods for cost and schedule growths? An ANOVA test was used in this instance to test for differences. Levene s test of homogeneity of variances was found not to be violated (p.05), which indicates that the population variances for each group of the procurement methods for cost and schedule growth were equal. The ANOVA, however, revealed that there were also no significant differences between procurement methods and cost growth, F(9,151) 0.217, p 0.05. Nor is there any significant difference in schedule growths, F(9,151) 1.553, p 0.05. The respondents were asked to estimate the direct cost of rework for the project they selected. The estimate provided by the respondents was then expressed as a ratio of cost growth so that the rework as a percentage of cost growth for each project could be determined. Fig. 6 displays a summary of direct rework costs as a percentage of cost growth (M 52.1%,SD 88.9%). While the mean cost growth for the projects was found to be 12.6%, rework contributed to 52.1% of the cost growth, with factors such as weather, industrial relations, and client/end-user change orders contributing to the remaining 47.9%. Clearly, reductions in rework significantly improve the cost performance of projects. Noteworthy, however, there were cases that experienced high direct rework costs 10% but came in under budget and experienced negative cost growths. In testing whether there was a significant relationship between cost growth and rework, Pearson s r correlation was computed Table 4. The correlation coefficients for the data revealed that cost growth and direct rework were significantly related, r 0.51, n 161, p 0.01, two tails and r 2 0.26 26%. Similarly, indirect rework was also significantly related to cost growth, r 0.25, n 161, p 0.01, two Table 5. Mean Schedule and Cost Growth for Different Procurements Procurement method Mean cost growth % Mean schedule growth % Traditional lump sum 15 23 Traditional cost plus 19 67 Traditional with provisional quantities 13 10 Design and manage 5 12 Construction management 11 25 Management contracting 6 6 Design and build 11 13 Novation 8 34 Turnkey/package deal 9 8 BOOT 9 2 Fig. 4. Percentage cost growth for projects JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002 / 23

Fig. 5. Percentage schedule growth for projects Fig. 7. Total rework costs as percentage of original contract value tails and r 2 0.06 6%, though it should be noted that Hinkle et al. 1988 recommend that a correlation of 0.70 to 0.90 be considered high and 0.50 to 0.70 moderate. This rule of thumb was recommended because the r-value is used to derive the value for r 2, which indicates that one variable can be predicted by changes in another. Thus, 26% of the variance in cost growth can be attributable to changes due to direct rework. The correlation analysis, presented in Table 4, also reveals that schedule growth and direct rework costs are significantly related, r 0.38, n 161, p 0.01, two tails and r 2 0.15 15%. However, the r 2 is low, which indicates that only 15% of the variance in scheduled growth can be predicted by changes due to direct rework. While this is a surprising finding, a detailed examination of the project data revealed that 27% of projects were delivered on or ahead of time despite experiencing cost increases due to rework. Once rework occurs, projects can be accelerated and resources reallocated to compensate for any delays that may arise, though this may be costly in dollar terms, especially if the rework influences the projects critical path. The occurrence of rework will invariably result in the contractors reevaluating their project schedules, as delays have the potential to lead to liquidated damages being incurred. If a delay occurs due to rework and the contractor is not responsible for it, then an extension of time or acceleration costs may be awarded, though this will depend on the type of delay and how it impacts the critical path. The reevaluation of a project schedule may well require project teams to work together to solve problems that may arise as a result of the rework. In doing so, they may identify alternative construction sequences and/or methods that can be used to complete the project on or ahead of time. Rework Costs The questionnaire survey asked the respondents to provide an estimate of the direct and indirect costs of rework that occurred in the project selected. Rework costs are very rarely, if ever, mea- Fig. 6. Rework as percentage of cost growth Fig. 8. Direct rework costs as percentage of original contract value 24 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002

Fig. 9. Indirect rework costs as percentage of original contract value sured by Australian construction organizations Love et al. 2000a, so the estimates provided by respondents were based on their cognizance of the project. The total rework costs were calculated by adding the direct and indirect estimates provided by the respondents. Fig. 7 identifies the mean and standard deviation of total rework costs for the 161 construction projects (M 12.0%,SD 13.56%). Figs. 8 and 9 display the distribution of the respondents estimates for direct (M 6.4%,SD 7.78%) and indirect (M 5.6%,SD 7.19%) rework costs. Though it was surprising to find that respondents considered indirect costs to be similar in amount to direct costs, especially as Love et al. 2000c have shown indirect rework costs can have a cost-multiplier effect as much as five times the actual direct cost of rectification. The total costs of rework have been found to vary considerably among projects. Some respondents reported rework costs to be less than 1% of a project s original contract value, while others have reported them to be as high as 80%. The degree of variability in the estimates given by the respondents suggests that many respondents may be unsure about the actual costs of rework incurred in the projects. The mean estimates of practitioners rework costs that were perceived to have been incurred in their selected projects are presented in Table 6. To test whether there were significant differences between the estimates of the respondents for rework costs direct and indirect an ANOVA was undertaken. The descriptive statistics revealed differences between the design consultants M 8.0%,SD 9.3), contractor (M 5.8%,SD 7.2%), and project manager (M 4.3%,SD 3.9%) in their estimates of direct rework costs. Levene s test of homogeneity of variances was violated ( p 0.05), which indicates that the population variances for each respondent type were not equal. The ANOVA revealed significant differences between respondents estimates of direct rework costs, F(2,158) 3.028, p 0.05. The results of Tukey s HSD post hoc test indicated that the differences identified in the estimates for direct rework costs were between design consultants and project managers (p 0.05). Consultant project managers typically act as the client s representative for projects and thus would possess reasonable knowledge to be familiar with the direct rework costs in the form of change orders and defects, as they invariably manage time and cost schedules. Project managers, however, may not be aware of the direct rework costs associated with redocumenting aspects of the project after clients have requested design changes and/or omissions. While client involvement in projects has been identified as a factor that can contribute to project success Walker 1994; Chan 1996; Love et al. 1998b, their involvement may also lead to rework. For example, drawing on the qualitative comments provided by the respondents, a project manager stated that the client was a decision-maker and actively involved in construction, resulting in scope and design changes throughout the construction. This in turn can lead to design consultants having to redocument or provide additional documentation, which can significantly affect their fee, as they are often not reimbursed for this service Tilley and McFallan 2000. Several design consultants articulated this point, with one stating that reworking of documentation is becoming a common occurrence on projects, which is not reflected in our fee. Another design consultant stated that a lot of rework had to be done on the documentation to reduce the scope of packages or substitute materials in an attempt to get the project within the budget. Redocumentation due to design changes and omissions initiated by clients and end-users appears to be a regular occurrence in Australian projects Love et al. 2000b; Tilley and McFallan 2000. Acknowledging this endemic practice, a design consultant stated Rework is an occurrence that consultants try hard to avoid because it can potentially lead to cost increases, which we are often not paid for. So the tendency, therefore, is to hold back on committing resources until all the information required to complete the task is available. Rework, nevertheless, often occurs and can usually be put down to poor planning or devoting of insufficient time to the planning and design before commencing construction. The allocation of resources and planning of the documentation process are important points that need to be addressed if rework is to be reduced Love et al. 2000a. Yet understanding why there are differences in the estimates of direct rework costs is a major research task in itself and thus worthy of further investigation. Table 6. Direct and Indirect Rework Cost Estimates by Respondent Type Respondent N Mean Deviation Direct rework costs Error Minimum Maximum Mean Deviation Indirect rework costs Error Minimum Maximum Design consultant 71 8.0 9.3 1.1 0.50 50.00 6.77 8.34 0.99 0.00 50.00 Contractor 53 5.8 7.2 1.0 0.10 30.00 5.46 6.87 0.94 0.00 30.00 Project manager 37 4.3 3.9 0.6 0.50 15.00 3.64 4.37 0.71 0.00 20.00 Total 161 6.4 7.8 0.6 0.10 50.00 5.62 7.18 0.56 0.00 50.00 JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002 / 25

Table 7. Direct and Indirect Rework Costs for Procurement Methods Used Procurement method N Mean Deviation Direct rework costs Error Minimum Maximum Mean Deviation Indirect rework costs Error Minimum Maximum Traditional lump sum 84 7.03 8.27 0.90 0.50 50.00 5.63 6.61 0.72 0.00 30.00 Traditional cost plus 1 20 20.00 20.00 10.0 2.76 10.00 10.00 Traditional with provisional quants 5 10 14.28 6.38 1.00 35.00 4.50 6.18 1.00 0.00 15.00 Design and manage 2 5.5 6.36 4.50 1.00 10.00 1.0 1.14 1.11 0.00 2.00 Construction management 26 4.11 4.15 0.81 0.50 15.00 4.48 5.67 1.65 0.00 20.00 Management contracting 1 2.0 2.00 2.00 0.0 0.00 0.00 Design and build 32 5.90 7.11 1.25 0.10 30.00 6.29 9.38 0.00 50.00 Novation 6 9.16 11.02 4.49 0.50 30.00 9.91 10.44 4.26 0.50 25.00 Turnkey and package deal 2 4.70 0.42 0.30 4.40 5.00 8.0 9.89 7.00 1.00 15.00 BOOT 2 1.5 0.70 0.50 1.00 2.00 2.0 1.41 1.00 1.00 3.00 Total 161 6.44 7.78 0.61 0.10 50.00 5.62 7.18 0.56 0.00 50.00 However, it would appear that consultant project managers might not fully understand how changes/omissions could affect the performance of design consultants particularly the way in which they manage the documentation process and as such explain the differences in the estimates for direct rework costs. The descriptive statistics revealed differences between the design consultants (M 6.77%,SD 8.34%), contractors (M 5.46%,SD 6.87%), and project managers (M 3.64%,SD 4.37%) in their estimates of indirect rework costs. Levene s test of homogeneity of variances was not violated (p 0.05), which indicates that the population variances for each group were equal. The ANOVA test revealed no significant differences between respondents estimates for indirect rework costs, F(2,158) 2.364, p 0.05. Noteworthy are the design consultants estimates that are almost twice as much as those estimated by project managers, which again demonstrates the variability associated with rework cost estimates, albeit not for the same projects. For both direct and indirect rework costs, contractors estimates are approximately midway between those of design consultants and project managers. This implies that contractors may well have a better understanding of actual rework as they invariably operate at the interface between design and construction. When respondents were asked to compare the rework costs of their selected project with others with which they had been involved, 12% stated that they were comparable to a very large extent, 16% to a large extent, 37% to some extent, 26% to a minor extent, and 9% not at all. The mode was found to be to some extent, and thus it can be concluded that the estimated rework costs reported are generally representative of industry practice in Australia. The rework costs found from this research appear to be relatively comparable to previous studies Davis et al. 1989; Burati et al. 1992; Abdul-Rahman 1995; Nylén, 1996; Josephson and Hammarlund 1999. However, such studies, with the exception of Barber et al. 2000, fail to differentiate or acknowledge the distinction between direct and indirect costs. Thus, the mean rework costs direct and indirect reported are considered benchmarks that can be used to pursue best practice in the Australian construction industry. In order to make practitioners aware of the actual costs of rework, a system is needed that can capture and identify rework, such as that developed by Burati et al. 1992, Low and Yeo 1998 and Love and Li 2000b, so that appropriate prevention techniques can be identified and implemented in future projects. Rework Costs and Procurement Methods Table 7 identifies the direct and indirect rework costs for each procurement method. Levene s test of homogeneity of variances was not violated for direct and indirect costs (p 0.05), which indicates that the population variances for each group were equal. The ANOVA revealed no significant difference between procurement methods for direct rework costs, F(9,151) 1.004, p 0.05 and indirect rework costs F(9,151) 0.624, p 0.05. No Tukey HSD post hoc tests were performed for both direct and indirect rework costs because at least one procurement method category had fewer than two cases. Consequently, procurement methods were reclassified into traditional and nontraditional. So, to determine whether there was a difference between traditional and nontraditional procurement methods for rework costs direct and indirect, at-test was undertaken. Table 8 presents the mean and standard deviation for total rework costs for traditional and nontraditional methods, and the results of the t-test are presented in Table 9. At the 95% confidence level, no significant difference in the total cost of rework was experienced in projects using different procurement methods. Therefore, it is concluded that rework costs do not significantly vary among procurement methods employed. Rework Costs and Project Types Refurbishment and renovation projects are considered prone to higher rework costs than new build projects because of the degree of uncertainty and complexity associated with the building work Table 8. Total Rework Costs for Traditional and Nontraditional Methods Procurement method N Mean deviation Error mean Traditional 89 12.77 13.50 1.43 Nontraditional 72 11.10 13.66 1.61 26 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002

Table 9. t-test for Procurement Methods and Total Rework Costs Levene s Test for equality of variances t-test for equality of means F Sig. t df Sig. 2-tailed Mean difference error difference 95% CI of difference Lower Upper Equal variances assumed 0.136 0.713 0.772 159.000 0.441 1.662 2.152 2.589 5.912 Equal variances not assumed 0.771 151.285 0.442 1.662 2.155 2.596 5.919 Table 10. Direct and Indirect Rework Costs for Project Types Project type N Mean Deviation Direct rework costs Error Minimum Maximum Mean Deviation Indirect rework costs Error Minimum Maximum New build 90 6.10 7.18 0.75 0.10 35.00 5.69 7.70 0.81 0.00 50.00 Refurbishment/Renovation 43 7.29 9.73 1.48 0.50 50.00 5.60 6.43 0.98 0.00 30.00 Fit-out 14 7.78 7.70 2.06 1.00 30.00 6.10 7.90 2.11 0.50 30.00 New build/refurbishment 11 4.95 4.67 1.41 0.50 15.00 5.81 5.92 1.78 0.00 20.00 Combination of all 3 3.33 1.52 0.88 2.00 5.00 0.66 0.577 0.33 0.00 1.00 Total 161 6.44 7.78 0.61 0.10 50.00 5.62 7.18 0.56 0.00 50.00 to be undertaken Love and Wyatt 1997. In fact, NEDO 1988 and Naoum and Mustapha 1994 indicate that facility types are linked to the concept of complexity and thus have influence on project performance. The direct and indirect rework costs for the various project types sampled in this research can be seen in Table 10. The ANOVA test was used to determine if there was a significant difference between project type and rework costs. Levene s test of homogeneity of variances for both direct and indirect rework cost was not violated (p 0.05), which indicates that the population variances for each project type were equal. The ANOVA test revealed no significant differences between project types for direct rework costs F(4,156) 0.489, p 0.05, indirect rework costs F(4,156) 0.371, p 0.05, and total rework costs, F(4,156) 0.824,p 0.05. Considering the results of the ANOVA test, it is concluded that rework costs do not significantly vary among project types. Conclusions The research presented in this paper set out to determine if there were significant differences in rework costs for different project types and procurement methods. Data from 161 Australian construction projects were obtained through a questionnaire survey. It was revealed that traditional lump sum methods are the most popular forms of procurement used in Australia, despite calls for the adoption of more integrated methods, such as design and build. Considering this finding, cost certainty an inherent feature of the traditional lump sum method appears to be a key driver for clients. This does not, however, preclude clients from demanding earlier completion as well. Similarly, pressures to reduce the time to design and produce contract documentation under this method can be compressed to meet client demands. These client demands may influence the quality of contract documentation produced, as errors and omissions may emerge that can result in rework and thus cause cost and schedule overruns. The cost and schedule growths for the projects sampled were calculated, and it was found that mean cost and schedule growths were 12.6 and 20.7%, respectively. While some projects recorded relatively high rework costs, others also experienced negative cost and schedule growths, which was an interesting and unexpected finding. This implies that the original contract period may have contained buffers, or items of work were deleted, or construction methods were altered or a combination of all so that the project could be delivered within budget and on time or less. The analysis revealed no significant differences between the cost and schedule growths on the projects sampled. In addition, cost and schedule growths were found to be significantly correlated with direct rework costs, which suggests that rework can adversely influence project performance. In examining this further, rework as a percentage of cost growth was calculated and found to amount to 52.1% of the total, so it can be concluded that rework can make a significant contribution to a project s cost overrun. Mean direct and indirect rework costs were found to be 6.4 and 5.6% of the original contract value, respectively. Nevertheless, rework costs were found not to vary significantly with project type and procurement method used. Yet there is a great need to reduce rework costs if projects are to improve their productivity and performance. In particular, the influences of an organization s management practices eg quality and learning and project management strategies on reducing rework costs should be examined. While this research has provided an insight into practitioners estimates of rework costs, the actual costs still remain relatively unexplored in construction. Acknowledgments The author would like to thank the three anonymous reviewers and Professor Mohan Kumaraswamy, Professor Albert Chan, Professor David Jaggar, Professor Gary Holt, Dr. Olusegun Faniran, and Jim Smith for their constructive comments, which helped improve this paper. JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002 / 27

References Abdul-Rahman, H. 1995. The cost of non-conformance during a highway project: A case study. Constr. Manage. Econom., 13, 23 32. Alreck, P. L., and Settle, R. B. 1985. The survey research handbook. Richard D. Irwin, Inc., Homewood, Ill. Ashford, J. L. 1992. The management of quality in construction. E&F Spon, London. Barber, P., Sheath, D., Tomkins, C., and Graves, A. 2000. The cost of quality failures in major civil engineering projects. Int. J. Quality Reliability Manage., 17 4/5, 479 492. Bowen, P. A., Hindle, R. D., and Pearl, R. 1997. The effectiveness of building procurement systems in the attainment of client objectives. Proc., CIB W-92 Procurement Symposium, Procurement The Key to Innovation, Univ. of Montreal, Montreal, 18th 22nd May, 39 49. Bresnen, M. J., Haslam, C. O., Beardsworth, A. D., Bryman, A. E., and Hell, T. E. 1988. Performance on site and the building client. Occasional Paper 42, Chartered Institute of Building, Ascot, U.K. Brookes, J. P. Jr. 1975. The mythical man-month: Essays on software engineering. Addison-Wesley, Reading, Mass. Burati, J. L., Farrington, J. J., and Ledbetter, W. B. 1992. Causes of quality deviations in design and construction. J. Constr. Eng. Manage., 118 1, 34 49. Chan, A. P. C. 1996. Determinants of project success in the construction industry of Hong Kong. PhD thesis, Univ. of South Australia, Adelaide, Australia. Chan, A. P. C., Yu, A. N. T., and Tam, C. M. 1999. Enhanced design and build An innovative system to procure a hospital project. Proc., CIB W-92 Procurement Symposium Harmony and Profit, January 25th 28th Chiang Mai, Thailand, 405 415. Construction Industry Development Agency and Masters Builders Australia CIDA. 1995. Measuring up or muddling through: best practice in the Australian non-residential construction industry. Sydney, Australia, 59 63. Cox, I. D., Morris, J., Rogerson, J. H., and Jared, G. E. 1999. A quantitative study of post contract award design changes in construction. Constr. Manage. Econom., 17 4, pp. Davis, K., Ledbetter, W. B., and Burati, Jr., J. L., 1989. Measuring design and construction quality costs. J. Constr. Eng. Manage., 115 3, 385 400. Evbuomwan, N. F. O., and Anumba, C. J. 1996. Towards a concurrent engineering model for design-and-build projects. Struct. Eng., 74 5, 73 78. Hanna, A., Russell, J. S., Gotzin, T. W., and Nordheim, E. V. 1999. Impact of change orders on the labor efficiency for mechanical contractors. J. Constr. Eng. Manage., 125 3, 176 184. Hashim, M. 1997. Clients criteria on the experience of procurement systems A Malaysian experience. Proc., of the CIB W-92 Procurement Symposium, Procurement The Key to Innovation, Univ. of Montreal, Montreal, 18th 22nd May, 273 284. Hester, W. T., Kuprenas, J. A., and Chang, T. C. 1991. Construction changes and change orders: Their magnitude and impact. Source Document 66, Construction Industry Institute, Univ. of Texas at Austin, Austin, Tex. Hinkle, D. E., Wiersma, W., and Jurs, S. G. 1988. Applied statistics for the behavioural sciences. 2nd Ed., Houghton-Mifflin, Boston. Hoedemaker, G. M., Blackburn, J. D., and Van Wassenhove, L. N. 1999. Limits to concurrency. Decision Sci., 30 1, 1 17. Holt, G. D., Proverbs, D., and Love, P. E. D. 2000. Survey findings on UK construction procurement: Is it achieving lowest cost, or value? Asia Pac. Building Construct. Manage. J., 5, 13 20. Ibbs, C. W., and Allen, W. E. 1995. Quantitative impacts of project change. Source Document 108, Construction Industry Institute, Univ. of Texas at Austin, Austin, Tex. Institute of Building. 1979. Contractor selection A guide to good practice. Estimating Information Service, 34, Autumn. Institute of Building, Ascot, U.K. Jahren, C. T., and Asha, A. M. 1990. Predictors of cost-overrun rates. J. Constr. Eng. Manage., 116 3, 548 552. Josephson, P.-E., and Hammarlund, Y. 1999. The causes and costs of defects in construction. A study of seven building projects. Autom. Constr., 8 6, 681 642. Knocke, J. 1993. Post construction liability and insurance. E&FSpon, London. Lahdenperä, P. 1995. Reorganising the building process: The holistic approach. VTT Publications, Technical Research Centre of Finland, Espoo. Lam, P. T. I., and Chan, A. P. C. 1995. Construction procurement systems in the Asia Pacific countries. Australian Project Manager, 16 2, 12 17. Love, P. E. D., Gunasekaran, A., and Li, H. 1998a. Concurrent engineering: A strategy for procuring construction projects. Int. J. Proj. Manage., 16 6, 375 383. Love, P. E. D., and Li, H. 2000a. Overcoming the problems associated with quality certification. Constr. Manage. Econom., 18 2, 139 149. Love, P. E. D., and Li, H. 2000b. Quantifying the causes and costs of rework in construction. Constr. Manage. Econom., 18 4, 479 490. Love, P. E. D., Li, H., and Irani, Z. 2000c. Construction rework: The multiplier effect in practice. 5th Int. Conf. on ISO 9000 & TQM School of Business, HK Baptist Univ. & Productivity & Quality Research Centre, National Univ. of Singapore, Singapore, 435 439. Love, P. E. D., Mandal, P., and Li, H. 1999. Determining the causal structure of rework in construction projects. Constr. Manage. Econom., 17 4, 505 517. Love, P. E. D., Mandal, P., Smith, J., and Li, H. 2000a. Modelling the dynamics of design error induced rework in construction. Constr. Manage. Econom., 18 5, 575 586. Love, P. E. D., Skitmore, R. M., and Earl, G. 1998b. Selecting a suitable procurement method for a building project. Constr. Manage. Econom., 16 2, 221 233. Love, P. E. D., Smith, J., Treloar, G., and Li, H. 2000b. Some empirical observations of service quality in construction. Eng. Constr. Architectural Manage., 7 2, 191 201. Love, P. E. D., and Wyatt, A. D. 1997. Communication and rework: Case studies of construction projects. CSIRO, DBCE DOC 97/38 (B), Australia. Low, S. P., and Yeo, H. K. C. 1998. A construction quality costs quantifying system for the building industry. Int. J. Quality Reliability Manage., 15 3, 329 349. Mackinder, M., and Marvin, H. 1982. Design decision-making in architectural practice. Research Paper 19, Institute of Advanced Architectural Studies, Univ. of York, U.K. Masterman, J. W. E. 1994. A study of the basis upon which clients of the construction industry choose their building procurement systems. PhD thesis, Univ. of Manchester Institute of Science and Technology, Manchester, U.K. Naoum, S., and Mustapha, F. H. 1994. Influences of the client, designer and procurement methods on project performance. Proc., CIB W-92 Procurement Systems Symposium, East Meets West, Department of Surveying, Univ. of Hong Kong, Hong Kong, 221 228. National Economic Development Office NEDO. 1988. Faster building for commerce. HMSO, London. Nylén, K.-O. 1996. Cost of failure in a major civil engineering project. Licentiate thesis, Division of Construction Management and Economics, Department of Real Estate and Construction Management, Royal Institute of Technology, Stockholm, Sweden. Oppenheim, A. N. 1992. Questionnaire design, interviewing, and attitude measurement. Pinter Publishers, London. Royal Institution of Chartered Surveyors RICS. 1996. Contracts in use: A survey of building contracts in use during 1995. Conducted by Davis Langdon and Everest, November, London. Sharif, A., and Morledge, R. 1997. The insensitivity of the procurement processes in the UK construction industry to the problems of the occasional buyers from the industry. Proc., CIB W-92 Procurement Symposium, Procurement A Key to Innovation, Univ. of Montreal, Montreal, Canada, 693 703. 28 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002

Tilley, P. A., and McFallan, S. L. 2000. Design and documentation quality survey designers perspectives. BCE DOC 00/113. CSIRO Building, Construction and Engineering, Melbourne, Australia. Turner, A. 1990. Building procurement. Macmillan, London. Walker, D. H. T. 1994. An investigation into the factors that determine building construction time performance. PhD thesis, Department of- Building and Construction Economics, Faculty of Environmental Design and Construction, Royal Melbourne Institute of Technology, Melbourne, Australia. Watkinson, E. D. 1992. Procurement alternatives. Faculty of Building Journal, Autumn. Nottingham, U.K. 4 6. Zeitoun, A. A., and Oberlander, G. D. 1993. Early warning signs of project changes. Source Document 91, Construction Industry Institute, Univ. of Texas at Austin, Austin, Tex. JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 2002 / 29